38104d3d31 · matheusbsbs/matheus-bagliano

@@ -1,10 +1,10 @@

1 To: git@vger.kernel.org

2 Bcc: lwn@lwn.net

3 -Subject: What's cooking in git.git (Aug 2016, #07; Thu, 18)

4 -X-master-at: d63263a4dee8fc7da9b97bbdedf9c0d1f33024d4

5 -X-next-at: 8e4bfbdedf0822fce7bdef0dcc37482a8deb3aca

3 +Subject: What's cooking in git.git (Aug 2016, #08; Wed, 24)

4 +X-master-at: 2632c897f74b1cc9b5533f467da459b9ec725538

5 +X-next-at: 3a084aaf40d3898679a48d19dac63e45ecde4cc7

6

7 -What's cooking in git.git (Aug 2016, #07; Thu, 18)

7 +What's cooking in git.git (Aug 2016, #08; Wed, 24)

8 --------------------------------------------------

9

10 Here are the topics that have been cooking. Commits prefixed with

@@ -12,8 +12,15 @@ Here are the topics that have been cooking. Commits prefixed with

12 '+' are in 'next'. The ones marked with '.' do not appear in any of

13 the integration branches, but I am still holding onto them.

14

15 -The preview for the upcoming 2.10 release has been tagged as

16 -v2.10.0-rc0.

15 +v2.10.0-rc1 was tagged last week, then I got sick and lost an entire

16 +day. I've merged a handful more topics to 'next' and marked 5 among

17 +them to be merged to 'master' by -rc2. A notable is the ja/i18n

18 +topic, which hopefully will reduce the translator's load somewhat.

19 +Also the change to arrange the file-descriptors to the tempfile be

20 +closed for subprocesses to avoid deadlocks on Windows is also in;

21 +even though it is not a new problem, the fix looked safe enough.

22 +Linus's "32-bit GPG key-ids are so last century" change is already

23 +in, as reported in the 2.10.0-rc1 announcement.

24

25 You can find the changes described here in the integration branches

26 of the repositories listed at

@@ -23,60 +30,62 @@ of the repositories listed at

30 --------------------------------------------------

31 [Graduated to "master"]

32

26 -* jk/tighten-alloc (2016-08-13) 2 commits

27 - (merged to 'next' on 2016-08-16 at 5399351)

28 - + receive-pack: use FLEX_ALLOC_MEM in queue_command()

29 - + correct FLEXPTR_* example in comment

30 -

31 - Small code and comment clean-up.

32 -

33 -

34 -* js/test-lint-pathname (2016-08-16) 1 commit

35 - (merged to 'next' on 2016-08-16 at d154f90)

36 - + t/Makefile: ensure that paths are valid on platforms we care

33 +* ab/hooks (2016-08-16) 1 commit

34 + (merged to 'next' on 2016-08-17 at b56e55d)

35 + + rev-parse: respect core.hooksPath in --git-path

36

38 - The "t/" hierarchy is prone to get an unusual pathname; "make test"

39 - has been taught to make sure they do not contain paths that cannot

40 - be checked out on Windows (and the mechanism can be reusable to

41 - catch pathnames that are not portable to other platforms as need

42 - arises).

37 + "git rev-parse --git-path hooks/<hook>" learned to take

38 + core.hooksPath configuration variable (introduced during 2.9 cycle)

39 + into account.

40

41

45 -* rs/mailinfo-lib (2016-08-13) 1 commit

46 - (merged to 'next' on 2016-08-16 at 14101e3)

47 - + mailinfo: recycle strbuf in check_header()

42 +* jk/difftool-command-not-found (2016-08-15) 1 commit

43 + (merged to 'next' on 2016-08-17 at 32baf03)

44 + + difftool: always honor fatal error exit codes

45

49 - Small code clean-up.

46 + "git difftool" by default ignores the error exit from the backend

47 + commands it spawns, because often they signal that they found

48 + differences by exiting with a non-zero status code just like "diff"

49 + does; the exit status codes 126 and above however are special in

50 + that they are used to signal that the command is not executable,

51 + does not exist, or killed by a signal. "git difftool" has been

52 + taught to notice these exit status codes.

53

54

52 -* sg/reflog-past-root (2016-08-15) 1 commit

53 - (merged to 'next' on 2016-08-16 at ee997a5)

54 - + t1410: remove superfluous 'git reflog' from the 'walk past root' test

55 +* lt/gpg-show-long-key-in-signature-verification (2016-08-16) 1 commit

56 + (merged to 'next' on 2016-08-17 at 1ee8a00)

57 + + Merge branch 'lt/gpg-show-long-key-in-signature-verification-maint' into lt/gpg-show-long-key-in-signature-verification

58 + (this branch uses lt/gpg-show-long-key-in-signature-verification-maint.)

59

56 - A small test clean-up for a topic introduced in v2.9.1 and later.

60 + "git log --show-signature" and other commands that display the

61 + verification status of PGP signature now shows the longer key-id,

62 + as 32-bit key-id is so last century.

63

64 + A last-minute merge of this topic was a bit scary but I made sure

65 + push-certificate codepath would not be negatively affected.

66

59 -* va/i18n (2016-08-12) 3 commits

60 - (merged to 'next' on 2016-08-16 at 6a06cd3)

61 - + t7411: become resilient to GETTEXT_POISON

62 - + t5520: become resilient to GETTEXT_POISON

63 - + t3404: become resilient to GETTEXT_POISON

67

65 - A handful of tests that were broken under gettext-poison build have

66 - been fixed.

68 +* lt/gpg-show-long-key-in-signature-verification-maint (2016-08-16) 1 commit

69 + + gpg-interface: prefer "long" key format output when verifying pgp signatures

70 + (this branch is used by lt/gpg-show-long-key-in-signature-verification.)

71

68 ---------------------------------------------------

69 -[New Topics]

72 + "git log --show-signature" and other commands that display the

73 + verification status of PGP signature now shows the longer key-id,

74 + as 32-bit key-id is so last century. This is based on older

75 + codebase, just in case somebody wants to have it.

76

71 -* ab/hooks (2016-08-16) 1 commit

72 - (merged to 'next' on 2016-08-17 at b56e55d)

73 - + rev-parse: respect core.hooksPath in --git-path

77

75 - "git rev-parse --git-path hooks/<hook>" learned to take

76 - core.hooksPath configuration variable (introduced during 2.9 cycle)

77 - into account.

78 +* rs/pull-signed-tag (2016-08-13) 4 commits

79 + (merged to 'next' on 2016-08-17 at cecca71)

80 + + commit: use FLEX_ARRAY in struct merge_remote_desc

81 + + merge-recursive: fix verbose output for multiple base trees

82 + + commit: factor out set_merge_remote_desc()

83 + + commit: use xstrdup() in get_merge_parent()

84

79 - Will merge to 'master'.

85 + When "git merge-recursive" works on history with many criss-cross

86 + merges in "verbose" mode, the names the command assigns to the

87 + virtual merge bases could have overwritten each other by unintended

88 + reuse of the same piece of memory.

89

90

91 * sb/checkout-explit-detach-no-advice (2016-08-15) 1 commit

@@ -89,62 +98,134 @@ of the repositories listed at

98 an explicit enough sign that the user knows what is going on. The

99 advice message has been squelched in this case.

100

101 +

102 +* tb/t0027-raciness-fix (2016-08-14) 1 commit

103 + (merged to 'next' on 2016-08-17 at 39a6635)

104 + + convert: Correct NNO tests and missing `LF will be replaced by CRLF`

105 +

106 + The t0027 test for CRLF conversion was timing dependent and flaky.

107 +

108 +--------------------------------------------------

109 +[New Topics]

110 +

111 +* cc/receive-pack-limit (2016-08-24) 3 commits

112 + - receive-pack: allow a maximum input size to be specified

113 + - unpack-objects: add --max-input-size=<size> option

114 + - index-pack: add --max-input-size=<size> option

115 +

116 + An incoming "git push" that attempts to push too many bytes can now

117 + be rejected by setting a new configuration variable at the receiving

118 + end.

119 +

120 + Will merge to 'next'.

121 +

122 +

123 +* hv/doc-commit-reference-style (2016-08-17) 1 commit

124 + (merged to 'next' on 2016-08-24 at b187d45)

125 + + SubmittingPatches: document how to reference previous commits

126 +

127 + A small doc update.

128 +

129 Will merge to 'master'.

130

131

95 -* lt/gpg-show-long-key-in-signature-verification (2016-08-16) 1 commit

96 - (merged to 'next' on 2016-08-17 at 1ee8a00)

97 - + Merge branch 'lt/gpg-show-long-key-in-signature-verification-maint' into lt/gpg-show-long-key-in-signature-verification

98 - (this branch uses lt/gpg-show-long-key-in-signature-verification-maint.)

132 +* js/no-html-bypass-on-windows (2016-08-19) 1 commit

133 + (merged to 'next' on 2016-08-24 at 05728f8)

134 + + Revert "display HTML in default browser using Windows' shell API"

135

100 - "git log --show-signature" and other commands that display the

101 - verification status of PGP signature now shows the longer key-id,

102 - as 32-bit key-id is so last century.

136 + On Windows, help.browser configuration variable used to be ignored,

137 + which has been corrected.

138

139 Will merge to 'master'.

140

141

107 -* lt/gpg-show-long-key-in-signature-verification-maint (2016-08-16) 1 commit

108 - + gpg-interface: prefer "long" key format output when verifying pgp signatures

109 - (this branch is used by lt/gpg-show-long-key-in-signature-verification.)

142 +* dg/document-git-c-in-git-config-doc (2016-08-23) 1 commit

143 + (merged to 'next' on 2016-08-24 at a923eb0)

144 + + doc: mention `git -c` in git-config(1)

145

111 - "git log --show-signature" and other commands that display the

112 - verification status of PGP signature now shows the longer key-id,

113 - as 32-bit key-id is so last century. This is based on older

114 - codebase, just in case somebody wants to have it.

146 + The "git -c var[=val] cmd" facility to append a configuration

147 + variable definition at the end of the search order was described in

148 + git(1) manual page, but not in git-config(1), which was more likely

149 + place for people to look for when they ask "can I make a one-shot

150 + override, and if so how?"

151

152 + Will merge to 'master'.

153

117 -* ak/curl-imap-send-explicit-scheme (2016-08-17) 1 commit

118 - - imap-send: Tell cURL to use imap:// or imaps://

154

120 - When we started cURL to talk to imap server when a new enough

121 - version of cURL library is available, we forgot to explicitly add

122 - imap(s):// before the destination. To some folks, that didn't work

123 - and the library tried to make HTTP(s) requests instead.

155 +* jk/format-patch-number-singleton-patch-with-cover (2016-08-23) 1 commit

156 + - format-patch: show 0/1 and 1/1 for singleton patch with cover letter

157

125 - Needs review and testing.

158 + "git format-patch --cover-letter HEAD^" to format a single patch

159 + with a separate cover letter now numbers the output as [PATCH 0/1]

160 + and [PATCH 1/1] by default.

161

162 + Will merge to 'next'.

163

128 -* bw/mingw-avoid-inheriting-fd-to-lockfile (2016-08-18) 2 commits

129 - - mingw: ensure temporary file handles are not inherited by child processes

130 - - t6026-merge-attr: child processes must not inherit index.lock handles

164

132 - The tempfile (hence its user lockfile) API lets the caller to open

133 - a file descriptor to a temporary file, write into it and then

134 - finalize it by first closing the filehandle and then either

135 - removing or renaming the temporary file. When the process spawns a

136 - subprocess after obtaining the file descriptor, and if the

137 - subprocess has not exited when the attempt to remove or rename is

138 - made, the last step fails on Windows, because the subprocess has

139 - the file descriptor still open. Open tempfile with O_CLOEXEC flag

140 - to avoid this (on Windows, this is mapped to O_NOINHERIT).

165 +* cp/completion-negative-refs (2016-08-24) 1 commit

166 + - completion: support excluding refs

167

142 - It was pointed out that not everybody has O_CLOEXEC flag during the

143 - review.

168 + The command-line completion script (in contrib/) learned to

169 + complete "git cmd ^mas<HT>" to complete the negative end of

170 + reference to "git cmd ^master".

171

145 - Waiting for the review discussion to settle.

146 - cf. <20160818173555.GA29253@starla>

172 + Waiting for a review.

173 +

174 +

175 +* ja/i18n (2016-08-24) 3 commits

176 + (merged to 'next' on 2016-08-24 at 3a084aa)

177 + + i18n: simplify numeric error reporting

178 + + i18n: fix git rebase interactive commit messages

179 + + i18n: fix typos for translation

180 +

181 + The recent i18n patch we added during this cycle did a bit too much

182 + refactoring of the messages to avoid word-legos; the repetition has

183 + been reduced to help translators.

184 +

185 + Will merge to 'master'.

186 +

187 +

188 +* jk/delta-base-cache (2016-08-23) 7 commits

189 + - t/perf: add basic perf tests for delta base cache

190 + - delta_base_cache: use hashmap.h

191 + - delta_base_cache: drop special treatment of blobs

192 + - delta_base_cache: use list.h for LRU

193 + - release_delta_base_cache: reuse existing detach function

194 + - clear_delta_base_cache_entry: use a more descriptive name

195 + - cache_or_unpack_entry: drop keep_cache parameter

196 +

197 + The delta-base-cache mechanism has been a key to the performance in

198 + a repository with a tightly packed packfile, but it did not scale

199 + well even with a larger value of core.deltaBaseCacheLimit.

200 +

201 + Will merge to 'next'.

202 +

203 +

204 +* js/cat-file-filters (2016-08-24) 4 commits

205 + - cat-file: support --textconv/--filters in batch mode

206 + - cat-file --textconv/--filters: allow specifying the path separately

207 + - cat-file: introduce the --filters option

208 + - cat-file: fix a grammo in the man page

209

210 + Even though "git hash-objects", which is a tool to take an

211 + on-filesystem data stream and put it into the Git object store,

212 + allowed to perform the "outside-world-to-Git" conversions (e.g.

213 + end-of-line conversions and application of the clean-filter), and

214 + it had the feature on by default from very early days, its reverse

215 + operation "git cat-file", which takes an object from the Git object

216 + store and externalize for the consumption by the outside world,

217 + lacked an equivalent mechanism to run the "Git-to-outside-world"

218 + conversion. The command learned the "--filters" option to do so.

219 +

220 +

221 +* sb/push-make-submodule-check-the-default (2016-08-24) 1 commit

222 + - push: change submodule default to check

223 +

224 + Turn the default of "push.recurseSubmodules" to "check".

225 +

226 + Alas, this reveals that the "check" mode is too inefficient to use

227 + in real projects.

228 + cf. <xmqqh9aaot49.fsf@gitster.mtv.corp.google.com>

229

230 --------------------------------------------------

231 [Stalled]

@@ -399,45 +480,44 @@ of the repositories listed at

480 --------------------------------------------------

481 [Cooking]

482

402 -* rt/help-unknown (2016-08-18) 2 commits

403 - - help: make option --help open man pages only for Git commands

404 - - help: introduce option --command-only

483 +* ak/curl-imap-send-explicit-scheme (2016-08-17) 1 commit

484 + - imap-send: Tell cURL to use imap:// or imaps://

485

406 - "git nosuchcommand --help" said "No manual entry for gitnosuchcommand",

407 - which was not intuitive, given that "git nosuchcommand" said "git:

408 - 'nosuchcommand' is not a git command".

486 + When we started cURL to talk to imap server when a new enough

487 + version of cURL library is available, we forgot to explicitly add

488 + imap(s):// before the destination. To some folks, that didn't work

489 + and the library tried to make HTTP(s) requests instead.

490

410 - Will merge to 'next'.

491 + Needs review and testing.

492

493

413 -* jk/difftool-command-not-found (2016-08-15) 1 commit

414 - (merged to 'next' on 2016-08-17 at 32baf03)

415 - + difftool: always honor fatal error exit codes

494 +* bw/mingw-avoid-inheriting-fd-to-lockfile (2016-08-23) 2 commits

495 + (merged to 'next' on 2016-08-24 at 8389b0e)

496 + + mingw: ensure temporary file handles are not inherited by child processes

497 + + t6026-merge-attr: child processes must not inherit index.lock handles

498

417 - "git difftool" by default ignores the error exit from the backend

418 - commands it spawns, because often they signal that they found

419 - differences by exiting with a non-zero status code just like "diff"

420 - does; the exit status codes 126 and above however are special in

421 - that they are used to signal that the command is not executable,

422 - does not exist, or killed by a signal. "git difftool" has been

423 - taught to notice these exit status codes.

499 + The tempfile (hence its user lockfile) API lets the caller to open

500 + a file descriptor to a temporary file, write into it and then

501 + finalize it by first closing the filehandle and then either

502 + removing or renaming the temporary file. When the process spawns a

503 + subprocess after obtaining the file descriptor, and if the

504 + subprocess has not exited when the attempt to remove or rename is

505 + made, the last step fails on Windows, because the subprocess has

506 + the file descriptor still open. Open tempfile with O_CLOEXEC flag

507 + to avoid this (on Windows, this is mapped to O_NOINHERIT).

508

509 Will merge to 'master'.

510

511

428 -* rs/pull-signed-tag (2016-08-13) 4 commits

429 - (merged to 'next' on 2016-08-17 at cecca71)

430 - + commit: use FLEX_ARRAY in struct merge_remote_desc

431 - + merge-recursive: fix verbose output for multiple base trees

432 - + commit: factor out set_merge_remote_desc()

433 - + commit: use xstrdup() in get_merge_parent()

512 +* rt/help-unknown (2016-08-18) 2 commits

513 + - help: make option --help open man pages only for Git commands

514 + - help: introduce option --command-only

515

435 - When "git merge-recursive" works on history with many criss-cross

436 - merges in "verbose" mode, the names the command assigns to the

437 - virtual merge bases could have overwritten each other by unintended

438 - reuse of the same piece of memory.

516 + "git nosuchcommand --help" said "No manual entry for gitnosuchcommand",

517 + which was not intuitive, given that "git nosuchcommand" said "git:

518 + 'nosuchcommand' is not a git command".

519

440 - Will merge to 'master'.

520 + Waiting for the review discussion to settle.

521

522

523 * po/range-doc (2016-08-13) 12 commits

@@ -460,25 +540,22 @@ of the repositories listed at

540 Waiting for the review discussion to settle.

541

542

463 -* tb/t0027-raciness-fix (2016-08-14) 1 commit

464 - (merged to 'next' on 2016-08-17 at 39a6635)

465 - + convert: Correct NNO tests and missing `LF will be replaced by CRLF`

466 -

467 - The t0027 test for CRLF conversion was timing dependent and flaky.

468 -

469 - Will merge to 'master'.

470 -

471 -

472 -* jk/diff-submodule-diff-inline (2016-08-16) 3 commits

473 - - diff: add SUBMODULE_DIFF format to display submodule diff

543 +* jk/diff-submodule-diff-inline (2016-08-23) 8 commits

544 + - diff: teach diff to display submodule difference with an inline diff

545 + - submodule: refactor show_submodule_summary with helper function

546 + - submodule: convert show_submodule_summary to use struct object_id *

547 + - allow do_submodule_path to work even if submodule isn't checked out

548 + - diff: prepare for additional submodule formats

549 - graph: add support for --line-prefix on all graph-aware output

550 - diff.c: remove output_prefix_length field

551 + - cache: add empty_tree_oid object and helper function

552

553 The "git diff --submodule={short,log}" mechanism has been enhanced

554 to allow "--submodule=diff" to show the patch between the submodule

555 commits bound to the superproject.

556

481 - Still being worked on.

557 + Waiting for the review discussion to settle.

558 + I think it is getting there.

559

560

561 * jk/reduce-gc-aggressive-depth (2016-08-11) 1 commit

@@ -535,15 +612,13 @@ of the repositories listed at

612 Any more comments?

613

614

538 -* mh/diff-indent-heuristic (2016-08-04) 8 commits

615 +* mh/diff-indent-heuristic (2016-08-23) 6 commits

616 - diff: improve positioning of add/delete blocks in diffs

540 - - is_blank_line: take a single xrecord_t as argument

541 - - xdl_change_compact(): keep track of the earliest end

542 - - xdl_change_compact(): fix compaction heuristic to adjust io

543 - - xdl_change_compact(): do one final shift or the other, not both

544 - - xdl_change_compact(): rename i to end

545 - - xdl_change_compact(): clarify code

546 - - xdl_change_compact(): rename some local variables for clarity

617 + - xdl_change_compact(): introduce the concept of a change group

618 + - recs_match(): take two xrecord_t pointers as arguments

619 + - is_blank_line(): take a single xrecord_t as argument

620 + - xdl_change_compact(): only use heuristic if group can't be matched

621 + - xdl_change_compact(): fix compaction heuristic to adjust ixo

622

623 Output from "git diff" can be made easier to read by selecting

624 which lines are common and which lines are added/deleted

@@ -552,10 +627,8 @@ of the repositories listed at

627 experiment to find a good heuristics.

628

629 Waiting for the review discussion to settle.

555 - cf. <cover.1470259583.git.mhagger@alum.mit.edu>

556 - cf. <20160804072705.a53mospcccksiz4e@sigill.intra.peff.net>

557 - cf. <20160804075631.jakbi5dbsbxsqcpr@sigill.intra.peff.net>

630 cf. <29c40cdd-363a-df09-f9e2-fe9070bb8a9c@ramsayjones.plus.com>

631 + cf. <a27aa17e-f602-fc49-92b3-2198e4772e47@ramsayjones.plus.com>

632

633

634 * cc/apply-am (2016-08-11) 40 commits

@@ -608,7 +681,6 @@ of the repositories listed at

681 We are almost there.

682

683

611 -

684 * jk/pack-objects-optim-mru (2016-08-11) 4 commits

685 (merged to 'next' on 2016-08-11 at c0a7dae)

686 + pack-objects: use mru list when iterating over packs

@@ -767,7 +839,21 @@ of the repositories listed at

839 would think.

840

841

770 -* pb/bisect (2016-08-12) 13 commits

842 +* pb/bisect (2016-08-23) 27 commits

843 + . bisect--helper: remove the dequote in bisect_start()

844 + . bisect--helper: retire `--bisect-auto-next` subcommand

845 + . bisect--helper: retire `--bisect-autostart` subcommand

846 + . bisect--helper: retire `--check-and-set-terms` subcommand

847 + . bisect--helper: retire `--bisect-write` subcommand

848 + . bisect--helper: `bisect_replay` shell function in C

849 + . bisect--helper: `bisect_log` shell function in C

850 + . bisect--helper: retire `--write-terms` subcommand

851 + . bisect--helper: retire `--check-expected-revs` subcommand

852 + . bisect--helper: `bisect_state` & `bisect_head` shell function in C

853 + . bisect--helper: `bisect_autostart` shell function in C

854 + . bisect--helper: retire `--next-all` subcommand

855 + . bisect--helper: retire `--bisect-clean-state` subcommand

856 + . bisect--helper: `bisect_next` and `bisect_auto_next` shell function in C

857 . bisect--helper: `bisect_start` shell function partially in C

858 . bisect--helper: `get_terms` & `bisect_terms` shell function in C

859 . bisect--helper: `bisect_next_check` & bisect_voc shell function in C

@@ -812,7 +898,7 @@ of the repositories listed at

898 with the more generic ref-filter API.

899

900 Rerolled.

815 - This also really needs review.

901 + Needs review.

902

903

904 * jc/merge-drop-old-syntax (2015-04-29) 1 commit

What's cooking (2016/08 #08)